A Data Structure for Real-Time Aggregation Queries of Big Brain Networks

Area of Science:

• Neuroscience
• Computational Biology
• Bioinformatics

Background:

• Neuroscience research increasingly relies on large-scale brain data from initiatives and consortia.
• Exploring gene-brain circuitry-behavior relationships necessitates fusing spatial connectivity data at multiple scales.
• Current manual data aggregation methods are time-consuming and inefficient for big connectivity data.

Purpose of the Study:

• To propose a novel data structure for interactive exploration of heterogeneous neurobiological connectivity data.
• To enable efficient aggregation queries for comparing multimodal brain networks at different scales.
• To facilitate the integration and exploration of experimental data within public data resources.

Main Methods:

• Development of a novel data structure optimized for disk access to handle large connectivity datasets.
• Implementation of aggregation queries for on-demand, real-time dissection of brain networks.
• Benchmarking against state-of-the-art graph engines for retrieving local brain area connectivity.

Main Results:

• The proposed data structure enables interactive exploration of neurobiological connectivity data with billions of edges.
• Aggregation queries allow real-time, spatially contextualized dissection of multimodal networks.
• The data structure outperforms existing graph engines in retrieving connectivity for user-defined brain areas.
• Demonstrated feasibility in analyzing fear-related neuroanatomy in mice and comparing autism-linked networks.

Conclusions:

• The novel data structure provides an efficient solution for handling and exploring big connectivity data in neuroscience.
• This approach facilitates cross-species comparisons, aiding in the selection of neural substrates for further study in model organisms.
• Enables researchers to embed and explore their own data within public resources without requiring extensive infrastructure.